LAYER 01 · DATABASE ENGINE
Time-series speed and lifecycle,
managed together by one core engine
Handle rapidly growing volumes of raw time-series data and relational business data with purpose-built engines. Scale the same data model from small edge devices to a single server or a cluster.
16.96M
TPCx-IoT Performance metric
200B+
Sensor Records per 1 TB
50–90%
Real-time compression efficiency
100+
Proven in the field at Korean companies
BUILT FOR PHYSICAL DATA
Process large volumes of AI and industrial data with confidence.
Optimized for high-frequency ingestion, infrequent updates, time-range queries, analysis focused on recent data, interval aggregation, and long-term retention.
01
High-volume data ingestion
append-optimized storage for Physical AI environments generating millions of data points per second
04
Schema flexibility
Efficiently accommodate high-cardinality tags and new metrics to respond to facility expansion.
02
Real-time analysis and monitoring
Sub-second queries deliver recent data instantly for dashboards, alerts and anomaly detection.
05
Data lifecycle
Reduce long-term storage costs by linking originals, rollups, backups, and archives into a single policy.
03
Time-based queries
Query from the last 5 minutes to multi-year trends with the same SQL using time-optimized indexing and aggregate functions.
06
Edge optimization
Reduces latency and cloud costs by filtering, compressing, and preprocessing on-site and transmitting only necessary information.
PURPOSE-BUILT DATA MODEL
Optimized table structures
for different types of data.
We do not put every workload into a single general-purpose table.
Choose engines that match your ingestion patterns, query methods, persistence needs and update frequency, then connect them with SQL.
TAG
Sensor and equipment time-series data
Store high-frequency measurements using tags and the time axis, and query long-term trends using ROLLUP.
APPEND · ROLLUP · Permanent
LOG
Events · Logs
Append events, packets and text logs with automatic arrival timestamps.
APPEND · Full text search · Permanent
TRANSACTION
Business and Status Data
Manage equipment status, inspection results and aggregated data with relational DML and explicit transactions.
FULL DML · ACID · Persistence
LOOKUP
Reference data · Metadata
Manage small, frequently referenced datasets, such as equipment codes and tag attributes, using a PRIMARY KEY.
UPDATE · JOIN · Persistent
VOLATILE
Real-time State and Cache
Quickly query and update current equipment state and session information in memory.
IN-MEMORY · UPSERT · Temporary
INGEST · INDEX · AGGREGATE · RETAIN
Automatically manage the industrial data life cycle.
High-speed APPEND
Handles continuous input loads of sensors, ticks, and logs using an APPEND path that bundles and transmits bulk rows.
Data Quality
Control duplicates and outliers during ingestion, and manage quality limits using LSL and USL.
Time & Tag Index
Use indexes tailored to time ranges, tag IDs and LOG keywords to locate the data you need in large raw datasets.
Retention Policy
It applies retention periods and storage policies according to the table lifecycle and automatically cleans up old data.
ROLLUP Family
Precompute long-term aggregates by second, minute, hour and day using basic, conditional, custom and JSON ROLLUP.
JSON Time Series
Retrieves and indexes JSON values of TAG and TRANSACTION by path, and rolls up numeric fields.
TRANSACTION TABLE
Use raw time-series data alongside business transactions.
JOIN high-frequency raw data in TAG and LOG tables with equipment master records, work orders, inspection results and alarm states. Manage business data that needs updating in TRANSACTION tables using familiar methods.
Transaction Control
BEGIN · COMMIT · ROLLBACK
Relational Model
PRIMARY KEY · UNIQUE · AUTO_INCREMENT
Full DML
INSERT · SELECT · UPDATE · DELETE
Query & Data
JOIN · INDEX · JSON · UPSERTBACKUP · MOUNT
How can services and apps use stored data right away?
NEXT LAYER
Explore Machbase Neo's data collection, transformation, API, visualization and automation features.
